Harnessing the Power of Event History, Spatial, and Network Analysis

• In Event: Methodological Pluralism

Abstract

Dependent data are frequently a feature of the event processes that social scientists seek to model. Conditional on covariates, units continue to exhibit dependencies due to factors such as geographic proximity and shared network ties. These dependencies present both complications and opportunities for social scientists seeking to understand inherently social event processes. This paper explores frontiers of event history research on spatial and network analyses and how survival modeling can benefit from approaches that incorporate and integrate both spatial and network approaches.

Despite their shared interest in dependent data, to date research on spatial and network survival models has largely developed on independent paths. This has led to different research foci in both sets of event history approaches. For example, spatial survival models often treat dependencies as a nuisance to be accounted for via frailty modeling approaches. Conversely, relational models in survival network analyses often seek to provide substantive explanations of processes of network tie formation.

This paper charts paths for a greater integration of these two sets of survival modeling approaches. It explores how the social selection models of network tie formation employed by de Nooy (2011) and other scholars can be advanced by incorporating spatial weighting functions examined by spatial analysts. The paper also examines how endogenous spatial dependencies can be incorporated more directly into survival modeling approaches rather than the current standard spatial frailty modeling approach. The paper also examines how approaches integrating spatial and network analyses, such as Hays, Kachi, and Franzese (2010), can be extended to the event history setting.

Author

• David Darmofal, University of South Carolina